HP8640B -20V Power Supply Down

8/16 2019

This beauty come with all protected caps, and the gear box is almost perfect, which was proven to be broken.  The band-switch for the frequency counter did miss contact, though.  The frequency counter band stay at 4-8 Mhz for ever.  

The FM/AM/Level Meter is crazy,  output did not reach to expected level.  Something wrong with it.  

The Failure Behaviour

I realized simple checking won't identify where the fault was. And blindly replace component won't help anything.

The extension board is even harder to obtain than the 8640B itself.  I decide provided a external testing bed for it. use a temporary regulating transistor power it from laboratory supply.

This -20V regulator supply is a strange design.  It use transistor helded some voltage from +30V voltage, and connected to system ground,  then get this -20 V regulator.  I may because lack of 'pure negative'  chip?  or use same design for all is cheap and easy.

I made a big mistake leading to deeply loss .  The tulum capacitor obviously easy to broken ,  so does the electrolytic one.  I disconnected them all, and for sure, they all in good status.  C7 is a Mica capacitor, which is realy faulty and always good quality, as i know.

The voltage is wiggling, not seems to be fixed at somewhere.  For get out the -20V, the Q4 should conducts harder, I suspected the Q2, Q4 and  all diode, but there all proved to be good.  VR3/VER4 VR5, good too.

I testing the voltage from pin 3 and pin 2 of U2 reference to -20V rail.   Pin3 is obviously high then pin 2, so the  output should be high enough to driven Q2/Q4 open, but it does not.  Pin6 voltage is seems too low to hardly driven Q4.

I even suspected the LM723 should had something bootstrapping circuit, make sure  the Q4  sure start.  Because the external reference VR3 is floating too, oh, my godness. 

I became hopeless, and ordered several LM732,and waiting delivery for days.  This beast frustrate me so deeply, changed the LM723 does not help anything.  And i almost desoldering everything in the board!  

I begin to feel like i was on the wrong way. I built a quick circuits to checking the LM723. And  all the LM723 is works every well, except on the HP board.

I fix it!

It's time to clear my mind.  changing everything won't help.  And no more blindness.   It's easy to become madness and lost judgement while things seems wired.  But i need make several rules for fix things,  I did some stupid replacement proved to be unnecessary. 

First one is, the resistor is strong ,  it maybe problematic but it can be identified with multimeter. Second, the transistor also strong,  and multimeter could identify the dead short and open lead(the diode or voltage on open lead could be shaky).   Of cause, the checking result is that nothing wrong with all the transistor and resistor.

There must be some way to identify the faulty point.   And I not follow the  PRINCIPLE OF A REGULATOR.  This regulator use feedback, first of all,  debugging circuits like this, should to checking the feedback path. 

The output voltage is wiggling, this means the output voltage could not be fixed, the feedback must be wrong somewhere.  The VR3 is ok, and R14, C5 is normal (by multimeter, and disconnected the C5).   How to make sure the R18 and R19,  R20 is ok?  A good idea, came up with,  I can short the R18 or R19, R20 that would told me something. 

Shorted the R19, R20, the output voltage changed, of cause, but the output still not stable, this means the R19, R20 not the faulty one. And, bingo, while the R18 shorted, the output is became rocky stable !!

I told you i made a bad presumption,  lost me deeply. I thought the mica capacitor never broken.  But i was so wrong.   Something wire is happened to C7, it's internal insulation is some kind of  broken, and leaky,  on board capacitance testing won't find that.  

I replaced C7 with a ceramic capacitor, and everything is back!

Story End!